Sensitivity experiments

STEP 1: run sensitivity experiment

You can run sensitivity experiments for the multiple soundings. For example, we can do experiments, in which the soil moisture to the wilting point (WILT) and the field capacity (FC)), as follows:

python $CLASS4GL/simulations/simulations_soil_moisture.py \
    --c4gl_path_lib $CLASS4GL \
    --station_id 74560 \
    --path_forcing $CLASS4GL_DATA/forcing/IGRA_PAIRS_20190515/ \
    --path_experiments $CLASS4GL_DATA/experiments/IGRA_PAIRS_20190515/ \
    --experiments " WILT FC " \
    --split_by 25 \
    --global_chunk_number 0

Note that another simulation script (simulations_soil_moisture.py) is used as before (simulations.py). You can have a look how the sensitivity simulations are defined by looking at the differences:

vimdiff  $CLASS4GL/simulations/simulations.py $CLASS4GL/simulations/simulations_soil_moisture.py

Please note that the different sensitivity experiments should be prepared in a consistent way as the base experiment. As such, if you have prepared you base experiment as an array job on a supercomputer according to this tutorial, please perform the sensitivity experiments in a similar way as provided in the next step.

STEP 2 (optional): sensitivity experiments on a supercomputer

You can also start the above experiments as array jobs for all the soundings over all the stations as follows:

ipython -i $CLASS4GL/simulations/batch_simulations.py -- \
  --c4gl_path_lib $CLASS4GL \
  --exec  $CLASS4GL/simulations/simulations_soil_moisture.py \
  --path_forcing  $CLASS4GL_DATA/forcing/IGRA_PAIRS_20190515/ \  
  --path_experiments  $CLASS4GL_DATA/experiments/IGRA_PAIRS_20190515/ \ 
  --experiments "WILT FC"

This will launch two array jobs. The entire batch of two experiments will be done on multiple parallel jobs distributed over two arrays.

STEP 3: open the sensitivity experiments with the data explorer

You can load the base experiment and the two soil-moisture sensitivity experiments with the data explorer:

ipython -i $CLASS4GL/interface/interface.py -- --c4gl_path_lib $CLASS4GL --path_forcing $CLASS4GL_DATA/forcing/IGRA_PAIRS_20190515/ --path_experiments $CLASS4GL_DATA/experiments/IGRA_PAIRS_20190515/ --experiments "BASE WILT FC"

You can now access the data (as described in the above steps) of all experiments:

c4gldata['BASE'] 
c4gldata['WILT'] 
c4gldata['FC']

In order to explore the different experiments, see steps 3 to 6 of the tutorial of the first experiment.

Next: Interactive data explorer.